Jump to content

Google to save us from pixelation


ittaku

Recommended Posts

Quote

What is this sorcery, you ask? Well, it’s Google combining two neural networks. The first one, the conditioning network, works to map the 8x8 pixel source image against other high-resolution images. Basically, it downsizes other high-res images to the same 8x8 size and tries to make a match on the features.

Alas, this will lead to the representation of VN genitalia as much larger than they actually are, considering how many of my own such photos are circulating throughout the Interwebs.

Link to comment
Share on other sites

5 minutes ago, ittaku said:

No need to take me seriously... For this to work google would need a massive bank of uncensored 2D hentai images which I doubt they have... or at least they won't admit to having, or give us public access to.

That was probably part of the 20% of the dark web that Anonymous shut down. 

Link to comment
Share on other sites

2 minutes ago, Narcosis said:

A dead end road, though. You can't create something out of nothing, especially in terms of ultra low resolution graphics.

Normally yes, but that's not what this is doing... it's trying to match from tiny amounts of data what is in its massive database to recreate the missing data.

Link to comment
Share on other sites

Just now, ittaku said:

Normally yes, but that's not what this is doing... it's trying to match from tiny amounts of data what is in its massive database to recreate the missing data.

It's not recreating data. It's basically guessing how the input should look like at a higher resolution - in this case, a resolution "high enough" to actually allow enough minimum detail. The end result is preety much a guess work and will vary with every single iteration. "Recreating" is a process of piecing the remains of something together to bring it back to it's former shape and look; if there's not enough data or it's missing, the end results might end in a falsification.

Link to comment
Share on other sites

Actually with a bit of manual cropping, minimal interpolation and some layer/mask magic I can reduce pixels in the pixelated region to 1/4 their size without any manual drawing, thus retaining only the original artist's work in a less pixelated form. Unlike redrawing the result still looks like part of the original and requires no drawing skills on my part. Here' s a rough example I slapped together for demonstration (obviously nsfw below):

It's pretty time consuming but kinda fun to do, and I guess I could improve a little bit on it by being more careful in my cropping. I have previously tried to automate the process but the manual input is crucial since they often overlay detail (usually bodily fluids...) within the pixelated regions.

Link to comment
Share on other sites

31 minutes ago, Ningen said:

 

Redrawing probably looks better and is faster than what you are trying there.

Well fortunately I do have someone redrawing for me on that very project. It was just a fun experiment. As I said, though, no redrawing will ever be the original artist's vision (if he/she had one and drew lots more detail at the time only to pixelate it, which isn't always the case.)

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...